Applied Energy 217 (2018) 496–508

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

A conceptual study on air jet-induced swirling plume for performance T

improvement of natural draft cooling towers
⁎
Yuanshen Lua, , Alexander Klimenkoa, Hugh Russella, Yuchen Daia, John Warnerb,
Kamel Hoomana
a
School of Mechanical and Mining Engineering, The University of Queensland, Qld 4072, Australia
b
Jord International, Sydney, NSW, Australia

H I G H L I G H T S

• The new concept of swirling plume significantly enhances cooling capacity of NDCTs.
• The cooling enhancement recovers power cycle output at high ambient temperature.
• The swirling plume creates an equivalent extra draft height above the tower.
• Air jet speed, direction, and nozzle size are the key controlling parameters.
• The concept consumes lower power than fan-forced coolers in long-term operations.

A R T I C L E I N F O A B S T R A C T

Keywords: In thermal power cycles including concentrating solar thermal (CST) plants, natural draft cooling towers
Natural draft cooling towers (NDCTs) are widely used heat-dumping facilities. One inherent drawback of NDCTs is that their cooling per-
Vortex cooling tower formance can be compromised by changes in ambient conditions, particularly temperature, which inevitably
Cooling enhancement reduces the net power output of the cycles. Current methods resolving this issue are limited in a few options
Swirling plume
including inlet air pre-cooling, exit air heating, and fan assistance, each with considerable operational or initial
Updraft vortex
Power cycle efficiency
cost. To more economically reduce energy efficiency losses of the power cycles due to inefficient cooling, this
paper proposes a new concept of swirling plume method for both dry- and wet-type NDCTs. The method is to
rotate the plume strongly like a tornado in the tower upper part and above the towers to increase the overall
tower updraft capacity (pressure). The swirling plume is induced by high-speed air jets distributed at certain
locations using a much smaller flow rate. A numerical investigation on a 20 m-tall dry-type NDCT model has
been conducted verifying that this concept increases the airflow and the water temperature drop of the heat
exchanger by at least 53.6% and 3.57 °C (39.2%), respectively, under 35 °C ambient temperature. This cooling
performance enhancement enables a half megawatt-scale sCO2-based CST power cycle to recover its net power
output, by 4.98%, to the level almost same as that at 30 °C ambient temperature. The air jet to create such a
swirling plume consumes only 1/7 of the recovered power roughly. Compared with a traditional fan-forced
cooler working under exactly the same condition, this concept requires significantly smaller energy in long-term
operations as it would run only during temperature extremes. A simplified analytical modelling has found that
the cooling tower performance is improved due to that the swirling plume creates an equivalent extra draft
height on top of the tower which is attributed to two different vortical effects. The overall angular momentum of
the swirl is a critical factor in these effects.

1. Introduction downstream to turbines is usually removed by these facilities. The heat

dump efficiency of the cooling towers is one of the crucial factors af-
Natural draft cooling towers (NDCTs) are widely used to remove fecting the overall power conversion (heat-to-electricity) efficiency of
heat in thermal power plants and many other industrial process. In the plants. Furthermore, the parasitic loss of power to run the cooling
Concentrated solar thermal (CST) power plants, the redundant heat equipment is an important consideration. While NDCTs have the

⁎
Corresponding author.
E-mail address: yuanshen.lu@uqconnect.edu.au (Y. Lu).

https://doi.org/10.1016/j.apenergy.2018.02.095
Received 12 December 2017; Received in revised form 9 February 2018; Accepted 12 February 2018
0306-2619/ © 2018 Elsevier Ltd. All rights reserved.
Y. Lu et al. Applied Energy 217 (2018) 496–508

Nomenclature Greek letters

A area (m2) α permeability in porous media (m2) or velocity distribution

AJn effective nozzle area normal to the jet stream, respectively factor
(m2) α∗, a ̂ constants in pressure-strain tensor
a jet direction angle (°) β, β∗, β ̂ constants in pressure-strain tensor
b jet direction angle (°) γ∗, γ ̂ constants in pressure-strain tensor
C inertial resistance factor in porous media ε characteristic factor of plume vortex for static pressure
Cp specific heat (J Kg−1 K−1) ηθ, ηz factor of conservation of momentum
Cμ constant in turbulent viscosity ηFt total efficiency of fan
dn thickness of porous media zone (m) ρ density (kg m−3)
H tower height (m) ρO, ρI air density outside and inside the cooling tower, respec-
hr convective heat transfer coefficient of radiator (heat ex- tively (kg m−3)
changers) (W m−2 K−1) μ, μe, μt laminar, effective, turbulent viscosity, respectively
K, Ke, Kt laminar, effective, turbulent thermal conductivity, re- (kg m−1 s−1)
spectively (W m−1 K−1) σω, σω2 constants in the transport equation of ω
Kto loss coefficient at the tower outlet ω turbulence energy specific dissipation rate (s−1)
KΣ total pressure loss coefficient throughout the cooling
tower Vectors
k turbulent kinetic energy (m2 s−2)
ma air mass flow rate (kg s−1) ê unit moment vector in co-ordinate directions
maJ jet air mass flow rate (kg s−1) M momentum vector
P pressure (pa) v velocity vector
Pr, Prt laminar, turbulent Prandtl number, respectively
ΔPr pressure difference across radiator (heat exchangers) (pa) Tensors/matrix
ΔPFt, ΔPFs total pressure and static pressure of fan, respectively (pa)
qr heat flux of the radiator (heat exchangers) (W m−2) Dij, Pij, Sij production tensors of Reynolds stresses
R cooling tower radius (m) δij indentity matrix
S volumetric source term εij dissipation tensor
T temperature (K) Φij pressure-strain tensor
U velocity component in x-, y-, or z- direction (m s−1)
VF volumic flow rate of fan (m3 s−1) Subscripts
v velocity (m s−1)
va, vF, vJ air velocity inside the cooling tower, at fan outlet, and of a air or air side
the jet, respectively (m s−1) E energy
vrJ, vθJ, vzJ air jet velocity in radial, tangential and vertical (axial) e effective
directions (m s−1) F fans
WF, WJ powers input of fans and jet nozzles, respectively (J s−1) I, O inside or inlet, outside or outlet
r, θ, z cylindrical system co-ordinates: radial, tangential, and J nozzle jet
axial (vertical) M momentum
x Cartesian system co-ordinates r radiator (heat exchangers)
z elevation (m)
i, j, k Cartesian co-ordinate serial

unbeatable advantage of using no power, one of their drawbacks is that performance of NDCTs against high ambient temperatures. For dry
the cooling performance on these towers, especially natural draft dry towers, various evaporation-based approaches have been proposed in-
cooling towers (NDDCTs), is highly compromised by changes in am- cluding dry-wet hybrid cooling [1–4] and inlet air precooling [5–7].
bient temperatures: the higher the ambient temperatures, the lower the These methods consume large volumes of water, and require accessory
heat transfer rate to atmosphere. Because of this, NDCTs are always water supply systems. They cannot be used in conjunction with wet
expected to be overdesigned to meet the desired cooling load at all cooling towers. One type of methods which is applicable to both dry
time. and wet towers is devising a way to increase the temperature of hot air
However, practical designs of NDCTs in thermal power plants are inside cooling towers to increase the air buoyancy so that the air flow
actually governed by a trade-off between the cooling performance and rate. This includes heat injection [8] and solar radiation [9] down-
the capital costs. As the result, they only provide sufficient cooling stream to the heat exchange zones (either heat exchangers or wet fills).
loads for target ambient temperatures selected based on, for example, Dynamical water distribution adjustment across the plane of heat ex-
the 95th–98th percentile of hourly ambient temperature in a year, not changers to optimise the local heat transfer is also considered a useful
100%. During the small portion of a year when ambient temperature is method in enhancing both cooling tower types [10,11]. In the cases
higher than the design values, the cooling towers cannot cool down the when crosswind presents, the cooling tower performance may be im-
cooling media sufficiently so that the overall power conversion effi- proved by taking use of the wind through windbreak walls, wind shells,
ciency drops. In base-load power plants, increasing fuel supply is a very deflectors, or the periphery of a tower base [12–23], which has been
common solution to offset the efficiency loss caused by poor cooling. A intensively studied in the past decade. More directly, air flow in NDCTs
better option is to recover the cooling capacity during the periods at a can be enhanced by deploying assistant fans at the inlets or above heat
less cost. exchange zones [24,25]. The method usually chooses options from
Currently, there are a few methods to improve the cooling super-big fans [26], lots of smaller fans [27,28], and less small fans with

497
Y. Lu et al. Applied Energy 217 (2018) 496–508

transition components to distribute airflow more evenly. However, in good agreement. Following the discussions on the controlling para-
most industrial fans work at air speeds much higher than those driven meters of the swirling plume, the paper estimates the power con-
by pure natural convection, unless they sacrifice their power efficiency. sumption of air jets and the benefit of the concept on the net power
This causes extra high pressure losses across heat exchangers which are output of a half megawatt-scale sCO2-based CST power cycle.
initially designed for significantly lower airflow rate, and thus the
overall efficiency of using fans are compromised. More importantly, the 2. CFD model setup
moving parts and the supporting structures of fans inevitably introduce
additional blockages to original cooling towers. The jet-induced swirling plume of a natural draft cooling tower was
On the other hand, rotational air flow, or air vortex, has been de- simulated numerically. A cylindrical natural draft dry cooling tower
monstrated by natural tornados to have the ability of flow augment. model with 20 m height and 12 m diameter was set up using the com-
Attempts to utilize this effect thus have been made on cooling towers/ mercial software ANSYS Fluent. The prototype of this model is an actual
chimneys through patents such as [29,30], despite lack of actual proof. 20 m-tall steel-membrane cooling tower built in a campus of The
Hemmasian Kashani et al. [31,32] numerically studied the influence of University of Queensland for a future small-scale CST power plant using
flow rotation in the over-shower zone of a natural draft wet cooling a supercritical CO2 Brayton cycle [34,35]. The tower body in the CFD
tower and found that the rotation intensifies heat and mass transfer in model only kept the main features of the prototype while most detailed
the zone [31] and increases the homogeneity of the flow field inside structures were ignored.
cooling towers under certain crosswind conditions [32]. The authors
proposed that the flow rotation could be generated by a type of de- 2.1. Models and boundary conditions
flectors, but failed to give any detail. The interest of the discussion was
also restricted in the tower internal flow field. Dobrego et al. [33] The boundary condition and dimensions of the cooling tower model
proposed to introduce a vortex flow of the vapour–air mixture above and the whole computational domain are illustrated in Fig. 2. As in-
spray zones in NDWCTs by setting water sprayers at a proper azimuthal dicated, eight air jet nozzles were proposed to be distributed evenly on
orientation. The vortex is believed to increase the operation stability of the tower inner wall at the location 10 m above the heat exchangers of
NDWCTs under medium and large wind loads. However, according to the cooling tower. The jet nozzles were modelled by eight curved rec-
the authors, only a small part (up to 30%) of the water spray mo- tangular faces with velocity inlet boundary conditions. The jet stream
mentum can be used to set the swirling vapour–air flow [33] and, more velocity were specified through both the magnitude and the direction.
importantly, the exact benefit of the method is still unknown due to the The latter was defined by azimuthal angle a and pitching angle b, which
simplicity of the study. was implemented in CFD through specifying radial, tangential, and
In view of the above, a new concept of swirling plume airflow boost axial components in a cylindrical coordinate system at the boundaries.
in NDCTs, be it wet or dry, is proposed by the authors. The basic idea is The temperature at these boundaries were set to be same as the aver-
to rotate the plume about the central vertical axis, like a tornado, in the aged one right above the heat exchangers, while the static pressure was
upper part and above the cooling towers by using a number of high interpolated during the iterations.
speed air jets distributed at certain locations. The tornado-like vortex In addition to the jet velocity, a jet area, Ajn, was also defined as the
increases the airflow through the heat transfer zones in lower part of nozzle area perpendicular to the jet stream. As the result, the areas of
the towers at the cost of a much smaller jet flow rate than that of the the curved rectangular faces varied from one case to another depending
main airstream. The jets can be produced by efficient fans/blowers on the jet direction, in order to keep Ajn constant at 0.089 m2, except for
through nozzles embedded in the inner surface of the tower wall at Case 8.
certain orientation angles, and may be sourced from the hot air above In this study, 9 different combination cases of jet magnitude, di-
the heat exchange or the ambient cool air, as shown in Fig. 1. rection, and area were considered with details as per Table 1 below.
The concept differs from the aforementioned existing proposals on These combinations cover the most typical setups in practical applica-
that it introduces the flow rotation actively, and adopts a much higher tions. Case 0 stands for the normal situation when the cooling tower
swirling ratio (defined as the ratio of circulation speed to the updraft operates only under pure natural draft mode with no swirling plume.
one) so that it can benefit from “extra bonus”—the increase of updraft Case 1 represents an arbitrarily selected configuration of the nozzle
force caused by the air vortex. Unlike conventional fan-assisted jets—angle a at 60° and angle b at 30° with the jet speed of 20 m/s.
methods, this concept eliminates visible moving parts and their sup- Cases 2–6 explores the effect of jet directions using the exactly same
porting structures, and thus has no influence to the NDCTs in normal size nozzle (0.089 m2) at the same speed (20 m/s) in Case 1, among
pure natural convection operations. It is ideal for intermittent use to which Cases 2–4 stands for air jets in radial direction only, in axial
cope with energy conversion efficiency losses of thermal power plants direction only, and in both radial and tangential directions, respec-
due to diurnal or seasonal changes in ambient temperature. tively. The selection of these cases enabled the determination of how
To the best of our knowledge, proposals similar to the concept of jet- the jet direction influence the overall effect of the swirling plume. The
induced swirling plume for natural draft cooling tower enhancement rest two cases (Cases 7 and 8) were used to examine the effect of jet
have not been discussed in open literatures yet, neither have the me- stream flow rate. Specifically, Case 7 differs from Case 1 in only the jet
chanisms of the vortex effect on tower draft force enhancement. In
fundamental fluid mechanics, theories connecting buoyant swirling Top view
turbulent plume and natural convective updraft effect are still under
Jet nozzles
development and have been seldom applied to practical engineering
Fans/blowers
problems so far.
To fill this gap, a numerical and analytical study on the jet-induced
swirling plume in a 20 m-tall NDDCT model has been conducted in this Heat exchange
zone (heat
paper. The effectiveness of the proposed concept on cooling tower
exchangers or
performance improvement is assessed and verified for the first time. wet fills)
Then, a simplified 1-dimensional model is introduced to preliminarily Air source
investigate the flow enhancement mechanisms on the NDDCT with the
air jet-induced swirling plume, revealing that there are multiple effects
attributed to the flow augment of swirling plume. Results from both the
Fig. 1. A schematic of the concept of the air jet-induced plume vortex on NDCTs.
numerical and analytical methods are cross-validated and found to be

498
Y. Lu et al. Applied Energy 217 (2018) 496–508

Whole domain Cooling tower 2.2. Governing equations

Pressure outlet
D H t = 20 m The Reynolds decomposed steady-state conservation equations of
70 m
D = 12 m mass, momentum, and energy for air are given by Eqs. (5)–(7)
Ht Hhx = 5 m
80 m Hs = 5 m ∂
Pressure inlet Wall (ρUj ) = 0
Hj = 15 m
∂x j (5)

Hs Hj ∂ (ρUj Ui ) ∂ ⎛ 2 ∂U ∂ ⎡ ⎛ ∂Ui ∂Uj ⎞ ⎤

Radiator + =− P + μ k⎞ + μ⎜ + ⎟−ρui uj
⎥ + SMi
∂x i ⎢ ⎝ ∂x j
⎜ ⎟

Porous media H ∂x j ∂x i ⎝ 3 ∂xk ⎠ ∂x i ⎠

Wall hx ⎣ ⎦
(6)
Velocity inlet
∂ ∂ ⎡⎛ Cp μt ⎞ ∂T ⎤
[ui (ρE + P ))] = k+ ⎜ ⎟

∂x j ∂x j ⎢
⎣⎝ Prt ⎠ ∂x j ⎥
⎦
Jet nozzles
vzj v
b ∂ ⎧ ⎡ ⎛ ∂Ui ∂Uj ⎞ ⎤⎫
vrj + Ui ⎢μe ⎜ + ⎟−ρui uj
⎥ ⎬ + SEi
Suction holes vșj ∂x j ⎨ ⎝ ∂x j ∂x i ⎠
⎩ ⎣ ⎦⎭ (7)
a ez
A jet nozzle er where ρui uj represents the averaged fluctuating turbulence stresses,
eș
Pressure outlet Reynolds stresses. Here, i,j,k = 1,2,3. E is the total energy and μe is the
effective viscosity which is the sum of the molecular and turbulent
Fig. 2. The boundary conditions in the CFD model. viscosities, i.e. μe = μ + μt . Reynolds stresses are second order tensors
composed of 9 unknown components, which need to be modelled to
speed (15 m/s), while Case 8 in only the nozzle size (0.3 m2). close Eqs. (5)–(7).
Each jet nozzle corresponds to a suction hole underneath. The Because of the anisotropic nature of the swirling turbulence, a
suction holes were modelled as pressure outlet boundaries with an higher order turbulence model—Reynolds Stress model is preferred
outflow rate same to the one of the jet nozzles. The actual static pres- [41]. Effectively, a separate transport Eq. (8) must be solved for each of
sure at the holes was then iteratively computed to match the flowrate. the Reynolds stress components. Since the air in this modelling is as-
The heat exchangers were modelled by a uniformly-thick zone sumed as incompressible Newtonian fluid, only six out of the 9 com-
treated as a porous media zone with its upper surface set as radiator ponents are actually independent and need to be solved, namely ρuu ,
boundary condition [36,37]. In the porous media zone, an additional ρvv , ρww , ρuv , ρuw , and ρvw .
source term Si representing the pressure loss within the heat exchanger
is added to the momentum equation. Si is expressed as ∂ ∂ ⎡ ⎛ μt ∂ui uj ⎤ ⎛ ∂Uj ∂U
(Uk ρui uj ) = + μ⎞ − ρu u + ρuj uk i ⎞ + Φij
⎥ ⎝ i k ∂xk
⎜ ⎟ ⎜ ⎟

∂xk ∂xk ⎢
⎣ ⎝ σk ⎠ ∂x k ⎦ ∂xk ⎠
C μe
SMi = −⎛ ρvi2 + vi⎞ dn (i = 1,2,3) −εij
⎝2 α ⎠ (1) (8)

where dn is the thickness of the porous zone. α and C are coefficients In the above equation, the pressure-strain tensor Φij , and the dis-
which were derived from an empirical pressure drop-air speed corre- sipation tensor εij need to be modelled using the expressions below:
lation provided by the manufacturer of the heat exchangers of the
2 1 1
Gatton cooling tower, namely: Φij = −C1 β ∗ρω ⎛−ui uj + kδij ⎞−a ⎛̂ Pij− Pkk δij ⎞−β ⎛̂ Dij− Pkk δij ⎞
⎝ 3 ⎠ ⎝ 3 ⎠ ⎝ 3 ⎠
ΔPr = 6.65va2 + 5.205va (2) 1
̂ ⎛Sij− Skk δij ⎞
−γ ρk
⎝ 3 ⎠ (9)
The radiator boundary calculates the heat flux qr added to the air to
representing the heat transfer at heat exchangers [38,39]:
2
εij = δij ρβ ∗kω
qr = hr (Tr −Tao) (3) 3 (10)

where Tao is the air outlet temperature of the heat exchanger. hr is the where the three additional tensors are defined as
∂Uj ∂U ∂U ∂U
overall heat transfer coefficient which can be estimated by another Pij = −ρui uk ∂x −ρuj uk ∂x i , Dij = −ρui uk ∂xk −ρuj uk ∂xk and Sij =
k k j i
empirical correlation of the prototype tower [40]: 1 ∂Uj ∂Ui k
(
2 ∂xi
+ ∂xj
). The turbulent viscosity is μt = Cμ ρ ω , and turbulence ki-
hr = 150.07va2 + 549.22va + 190.95 (4) 1
netic energy is k = 2 ui ui . The constants are α∗ = 1, β ∗ = 0.09 , α ̂ = 0.775,
Inlet and outlet pressure boundaries were applied to the side and the β ̂ = 0.196 , Cμ = 0.09, γ ̂ = 0.495, Pr = 0.74 , and Prt = 0.85.
top of the computational domain to define the ambient air condition, The transport equation of Reynolds stresses Eq. (8) involves another
where the static pressure and the temperature were specified as zero unknown quantity ω , turbulent frequency, which requires the equation
and 35 °C, respectively. below to close the whole set of governing equations:

Table 1
Key parameters for the air jet nozzles in all 9 cases for 35 °C ambient temperature.

Parameter Case 0 Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8

a (°) N/A 60 0 N/A 60 70 60 60 60

b (°) 30 0 90 0 30 45 30 30
v (m/s) 20 20 20 20 20 20 15 20
Ajn (m2) 0.089 0.3

499
Y. Lu et al. Applied Energy 217 (2018) 496–508

∂ ∂ ⎡⎛ μ ∂ω ⎤ α∗ω ∂uj reduces the air flow rate by around 14.5% compared with the baseline
(Uj ρω) = ⎜ μ + t⎞ ⎟ − ρui uj −βρω2 case simply due to the increase of the ambient temperature from 30 °C
∂x j ∂x j ⎢
⎣⎝ σ ω ⎠ ∂x j
⎥ k
⎦ ∂x i
to 35 °C. The decreased water temperature reduces following the
1 ∂k ∂ω
+ 2(1−F1) ρ weakening of the cooling tower’s heat dumping capacity, because of not
σω2 ω ∂x j ∂x j (11) only the smaller flow rate but also a less ITD (initial temperature dif-
where β = 0.075, σω = 2, σω2 = 1.168 and F1 is the blending function ference) between the water and the air. However, with the assistance of
[42]. the air jet-induced swirling plume as Case 1, the air flow rate and
The spatial discretization for above governing equations uses QUICK consequently the decreased cooling water temperature in the cooling
scheme and the pressure-velocity coupling method adopts the pressure- tower is remarkably improved. Specifically, the two indicators are in-
based segregated algorithm SIMPLE (Semi-Implicit Method for creased compared with those of Case 0 by 52.2% and 38.6%, respec-
Pressure-Linked Equations). The entire CFD domain was discretised by tively. The latter is nearly recovered to the design point, i.e. 12.71 °C
over 7.2 million structural mesh cells, prior to which a test for grid compared with 12.60 °C. The air jet mass flow rate 16.5 kg/s accounts
independence was done. Refinements of the cell size adjacent to the for only 15.2% of the total airflow in Case 1.
tower inner wall and in high-gradient regions such as the radiator and The swirling plume is developed above the air jet nozzle plane in
nozzles were made through setting the first layer thickness to 7 mm Case 1 as expected, while in the tower space upstream the nozzles, the
with the aspect ratio no more than 4. A growth rate of 1.05 was then air flow keeps flowing upward uniformly just as that of Case 0. The
applied to the rest of the cells. A good overall mesh quality was tornado-like vortex does remain when the plume leaves the tower
achieved with the maximum orthogonal skewness of 0.33 and the outlet although the streamlines gradually become less spiral with the
maximum aspect ratio of 8.6. The solutions were taken as converged increase of the height. The development of the vortical plume along
when all scaled residuals of the dependent variables dropped to the with the height can be analysed through plotting the plume velocity
order of 10−5 and the additionally monitored variables remained in- profiles at different elevations. Fig. 5 depicts the averaged axial and
variable. tangential plume velocity components versus radius for the Cases 0–4 at
the elevations of y = 5.5 m, 18 m, 20 m, 25 m, 30 m, and 40 m, re-
spectively. The radius was defined as the horizontal distance between a
2.3. Model validation point and the vertical axis of the tower.
Without the induced swirl in Case 0, the axial velocity profile ex-
As the conceptual air jet-induced swirling plume has not been yet hibits the expected transition from “top hat” shape to “Gaussian” shape.
realised on the cooling tower prototype, there is no experimental data The momentum flux of the plume above the cooling tower is not con-
available to be used for validation of the CFD results with the swirling served because of the presence of buoyancy, and thus the axial velocity
plume. However, the results without plume vortex can be compared increases with height until the plume is fully established [43]. The
against existing experimental data of the Gatton NDDCT by setting the tangential velocity of the plume is nearly zero everywhere as there is no
simulation conditions same as those of the in-situ measurements. flow rotation.
In the past a few years, a series of tests have been conducted on the Nonetheless, the velocity profiles in Case 1 appear completely dif-
Gatton NDDCT including the “constant heat rejection test” [35,40]. In ferent. The tangential velocity profile presents a sharp peak in the near-
this test, a constant heat was supplied to the cooling tower by an oil wall region at the height of 18 m, just above the jet nozzle plane. This
fired heater at different ambient temperatures. The air-side and tube- suggests that the rotating motion only exists in a ring zone adjacent to
side temperatures of the heat exchangers were measured and then the inner tower wall when the plume just encounters the jet. The
plotted against the ambient temperature. In the CFD model, the same swirling ratio at this plane is around 6, which is much higher than the
constant heat flux were applied while the atmospheric temperature was ones in [31]. Along with the rise of the plume, the rotation diffuses
varying in the same range. Fig. 3 shows the air exit temperature of the inward (toward the tower axis) or both inward and outward, while the
cooling tower obtained from both the experiment and the CFD simu- maximum velocity magnitude decreases. Meanwhile, the position of
lations at a constant heat load of 845 kW with a certain water mass flow tangential velocity peak gradually shifts to the tower axis, suggesting an
under different ambient temperatures. The comparison in the figure evolution of the original swirl toward an irrotational (free) vortex with
indicates a good agreement between both the results suggesting the a rotational core in the centre. There is a remarkable increase of
CFD modelling method is reliable. magnitude in axial velocity profile at the plane of y = 5.5 m comparing
Case 1 with Case 0, while both are flat. This directly indicates the air
3. CFD results on the NDDCT performances with swirling plume mass flow rate through the heat exchangers in the former is increased
without causing any disturbance at this height. The axial velocity
All the 9 cases were simulated by assuming the ambient tempera- profiles at elevations above tower outlet are much wider in Case 1,
ture as 35 °C without the existence of crosswind, and the radiator
temperature was fixed at a selected number 51.5 °C while the heat flux 50
was left variable. This condition represents the scenario that the up-
45
Air exit temperature at heat

stream temperature entering the cooling tower is required to be

maintained constant. In the following discussions, the air mass flow rate
40
exchanger (°C)

in kg/s and the decreased cooling water temperature (°C) in °C of the

heat exchanger are the main metrics used to assess the “cooling per- 35
formance” of the cooling tower.
The cooling performances of the NDDCT model is firstly assessed for 30
Measured data
the Cases 0–4, as summarised in Table 2. Here, for convenience, the
25 CFD results
water temperature change was derived based on assuming a fixed water
flow rate in the heat exchanger. The design point of the cooling tower
20
prototype of this model is also listed in the table as a baseline for the 10 12 14 16 18 20 22 24 26 28 30
comparison. The steady state 3D air streamlines in these five cases are
Ambient temperature (°C)
shown in Fig. 4.
The cooling performance enhancement due to the swirling plume is Fig. 3. The experimental and numerical air exit temperature of the cooling tower at an
845 kW heat load.
simply observed in the comparison between Case 0 and Case 1. Case 0

500
Y. Lu et al. Applied Energy 217 (2018) 496–508

Table 2
Comparisons in cooling performance among Cases 0, 1, 2, 3, and 4 as well as the baseline.

Parameter Case 0 Case 1 Case 2 Case 3 Case 4 Design point

Ambient temperature (°C) 35 35 35 35 35 30

Air mass flow rate (kg/s) 71.38 108.69 51.37 73.14 96.153 83.45
Decreased temperature of the cooling water (°C) 9.09 12.60 7.21 9.26 11.42 12.71
Air face velocity at the heat exchanger (m/s) 0.55 0.84 0.40 0.85 0.74 0.64
Air temperature at the heat exchanger exit (°C) 43.22 42.45 41.67 43.08 42.74 39.82
Overall UA based on the face area of the heat exchanger 538.40 758.13 434.64 766.16 679.51 603.89
Total air jet mass flow rate (kg/s) N/A 16.5 16.5 16.5 16.5 N/A
Percentage changed in flow speed, based on Case 0 N/A 52.3% −28.0% 2.47% 34.71% N/A
Percentage increased in heat transfer rate, based on Case 0 N/A 38.7% −20.7% 1.93% 25.7% N/A
Leverage factor of plume vortex N/A 2.26 −1.21 0.11 1.50 N/A

which implies that the hot plume out of the cooling tower spreads out as those in Case 1.
more at the same height with the presence of the swirl, which is con- According to Figs. 4 and 5, the plume vortex appears only when the
sistent with the observation in the temperature contours (Fig. 6). jet stream has a tangential component. A jet with only a radial mo-
The reason for the increased air flow rate inside the cooling tower mentum component does not increase the overall airflow in the cooling
may be understood through an examination of the static pressure var- tower, as verified by Case 2. More precisely, it adversely affects the
iations. Fig. 7 shows static pressure contours at the mid-xy plane of the performance by causing disturbance to the main flow (draft) in the
NDDCT model. Fig. 7a suggests that the static pressure inside a cooling upper part of the cooling tower causing mixing losses only. Con-
tower is always smaller that outside the tower at the same elevation. sistently, the axial velocity at y = 5.5 m exhibits a lower profile com-
The difference maximises just above the heat exchanger and fades away pared to Case 0. In Case 3, on the other hand, the purely vertical air jet
along the tower. It is the maximum pressure difference that determines results in an increased overall cooling tower airflow. However, the in-
the airflow across the heat exchanger. In pure natural convection case crease is quite minor compared with Case 0. The axial velocity profiles
(Case 0), the static pressure difference reduces to zero slightly above the at y = 18 m and 20 m clearly indicate that the main convective airflow
tower outlet. However, with the existence of the plume vortex in Case 1, of the cooling tower does not significantly benefit from the jet mo-
the negative pressure zone is extended to a considerable height above mentum.
the tower exit. And, the maximum pressure difference across the heat Case 4 proves that the jet combining radial and tangential mo-
exchangers is larger than that of Case 0, as if the tower wall is extended mentum components is also able to enhance the cooling tower perfor-
further up. Quantitatively, Fig. 8 shows the variations of the static mance. In this situation, the air jet is more effective than that of Case 3,
pressure with the elevation from the ground to 70 m for the two cases. but less effective than that of Case 1. It is interesting to note that the
The pressure at one elevation was averaged over circular horizontal axial velocity component at the height slightly above the jet nozzle
face with the radius same as the tower’s. As seen, Case 1 has a total plane, namely y = 18 m, shows a clear “down draft” near the cooling
equivalent draft pressure of 9.0 Pa, almost double of that in Case 0. tower wall. This probably affect the enhancement of the main airflow.
For a more comprehensive understanding of the underlying physics, Table 3 lists the cooling performance in Cases 1 and 5–8. Compared
we extended the simulations to three more cases. As mentioned above, with Case 1 as our benchmark, Case 5 bends the jet stream 10° closer to
Cases 2–4 represent the cases with air jets in radial direction only, in the tower wall, i.e. increase the azimuthal angle a to 70°, while the
axial direction only, and in both radial and tangential directions, re- pitching angle b remains unaltered. This increases the benefit of the
spectively. The air jet speed and flowrate in the 3 cases are all the same vortex. On the contrast, Case 6 is worse than Case 1, which only

Fig. 4. 3D streamlines starting from the heat exchanger for Case 0 (a), Case 1 (b), Case 2 (C), Case 3 (d) and Case 4 (e).

501
Y. Lu et al. Applied Energy 217 (2018) 496–508

4 4
Tower axis Case 0 Case 0
3.5 3.5
3 3

Air speed (m/s)

Air speed (m/s)

2.5 2.5
2 2
Tower wall
1.5 1.5
1 1
0.5 0.5
0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Radius (m) Radius (m)

5 5
Case 1 Case 1
4.5 4.5
4 4
Air speed (m/s)

Air speed (m/s)

3.5 3.5
3 3
2.5 2.5
2 2
1.5 1.5
1 1
0.5 0.5
0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Radius (m) Radius (m)

3 3
Case 2 Case 2
2.5 2.5
Air speed (m/s)

Air speed (m/s)

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Radius (m) Radius (m)

3 3
Case 3 Case 3
2.5 2.5
Air speed (m/s)

Air speed (m/s)

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Radius (m) Radius (m)
Fig. 5. The averaged axial (left) and tangential (right) velocity components versus the radius for Cases 0–4 at the elevations as indicated in the legend. The origin point of x-axis
represents the tower axis.

increases the angle b to 45°. (Case 8) will cause a further increase in the total airflow rate through
Comparisons between Cases 1, 6, 7, and 8 are made to examine the cooling tower.
whether the jet speed and the normal nozzle area have any effect on the In Tables 2 and 3, the leverage factor of vortex is defined as the ratio
effectiveness of the jet-induced vortex in the cooling tower. This com- of the increased overall air mass flow rate based on Case 0 to the total
parison eliminates the influence of the jet direction. Case 7 indicates air jet mass flow rate. It indicates the improved air flow rate with in-
that reducing the jet speed, compared to that of Case 1 while main- ducing air jet for a given configuration. The quantity has a practical
taining the nozzle dimension, results in decreased cooling performance. meaning as it is related to the overall energy efficiency of the jet-in-
On the other hand, increasing the nozzle area at the same jet speed duced vortex system in a NDCT if the jet is generated through fans or

502
Y. Lu et al. Applied Energy 217 (2018) 496–508

5 5
Case 4 Case 4
4.5
4 4

Air speed (m/s)

3.5

Air speed (m/s)

3 3
2.5
2
2
1.5
1
1
0 0.5
0 2 4 6 8 10 12 14 16 18 20 0
-1 0 2 4 6 8 10 12 14 16 18 20
Radius (m) Radius (m)

Fig. 5. (continued)

Fig. 6. Temperature contours at the mid-xy plane for Case 0 (a) and Case 1 (b).

Fig. 7. Static pressure contours at the mid-xy plane in Case 0 and Case 1. For convenience, the static pressure here is redefined as a gauge pressure without consideration of elevation
head, namely, by assuming the elevation head zero everywhere.

503
Y. Lu et al. Applied Energy 217 (2018) 496–508

80

70 Case 0 Case 1

60 Streamline
Elevation (m)

50

40 Heat exchanger
Equivelant extended draft height—ǻH
30
Ground
20 Without swirling plume With swirling plume
Visable draft
10 height—H=15 m Fig. 9. An ideal model of NDDCT with and without the air jet-induced swirling plume.
0
1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10
α 1
Static pressure (Pa) (P1−P2)−ρ2 gz 2 = ρ va2 + Kρa va2
2 2 2 (13)
Fig. 8. The mean air static pressure versus the elevation for Cases 0 and 1.
It can be expected that pressure P2 is slightly smaller than that of the
ambient air at same height far away from the cooling tower (position
blowers. 3), and the small difference is just a dynamic pressure loss due to the
airflow exists from the tower outlet [25], i.e.:
4. Analysis and discussions α
P2 = P3− Kto ρ2 va2
2 (14)
4.1. A Simplified analytical model for NDDCT performance with swirling
plume where Kto is the loss coefficient at the tower outlet.
Eq. (14) is then substituted into Eq. (12) to eliminate P2, which
The analysis below was carried based the above NDDCT model to results in three dynamic pressure terms expressed as functions of va.
explore the mechanism under the airflow enhancement by the swirling These terms can be combined into a single dynamic pressure term
K + α (1 − Kto)
plume. Fig. 9 shows an ideal cylindrical NDDCT model with or without 2
ρa va2 . Meanwhile, (P1−P3) is the difference between ambient
air jet nozzles in a 2D view. static pressures at heights 1 and 2, and thus is equal to ρ1 gz 2 . Therefore,
In with absence of plume swirl, NDCTs work on the natural con- the following equation can be obtained:
vection effect (stack effect): driven by the net buoyancy force, air he- K + 1−Kto 2
ated by the heat exchangers is less dense so that it moves upward. Cool ρ1 gz 2−ρ2 gz 2 ≅ (ρO −ρI ) gH ≅ ρI va
2 (15)
air underneath is thus sucked in to replace the hot one, forming a
steady-state air flow. If one considers an arbitrary streamline in the where H is the cooling tower height which is assumed to approximate to
cooling tower as shown in Fig. 9, the air accelerates from position 1 at altitude z2. According to [16], α ≈ 1 for dry cooling towers where the
the ground level, overcomes all the resistances in the tower and finally heat exchangers are arranged horizontally near the tower base.
reaches position 2 at the outlet at a speed. In this process, it can be Eq. (15) can be thought of as the so-called Draft Equation for natural
approximated that the mechanical energy along the streamline is con- draft cooling towers [25]. As the air flow inside the cooling tower is
served, i.e.: isentropic and continuous, va remains constant anywhere in the above
tower model, regardless of the boundary layer at the tower wall. Then,
1 1 the total air mass flow rate inside the tower, ma, is expressed as
P1 + ρ v12 + ρ1 gz1 = P2 + ρ2 v22 + ρ2 gz 2 + ΣHL
2 1 2 (12) ma = ρI va A , where A is the cross-section area of the tower.
In Eq. (12), P and z are static pressure and height respectively. ρ1 If one considers the air momentum at the tower outlet face, it can be
and ρ2 are equivalent to air densities outside and inside the cooling expressed in a cylindrical coordinate system as
tower, namely ρo and ρi . ΣHL denotes the sum of all flow resistances M = Mr er̂ + Mθ eθ̂ + Mz eẑ = ma v (16)
along the flow direction. The elevation at position 1 is zero while the
velocity on the streamline, far away from the tower inlet, is negligibly One argues that the momentum in the vertical direction, i.e. z
small. If use va as the face mean velocity at position 2 instead of v2, a component, is the dominant component and Mz is calculated as:
1 1
velocity distribution factor α is introduced, namely 2 ρ2 v22 = α 2 ρ2 va2 . As 2A (ρO −ρI ) gH
mentioned above, the mean velocity va is the same at any height in the Mz = ma va ≅
KΣ (17)
tower between the heat exchanger and the outlet. It is more convenient
1
to express ΣHL in the form of dynamic pressure of air, i.e. ΣHL = 2 Kρa va2 , Now let us consider the situation when the air jet-induced swirling
where K is the pressure loss coefficient between positions 1 and 2, and plume is applied in the cooling tower. As the jet nozzles source the air
ρa is the mean air density between ρ1 and ρ2 , or ρO and ρI . inside the cooling tower, mass is conserved in the airflow. Similarly,
Therefore, Eq. (12) can be rearranged: one can select a random streamline through the cooling tower that is

Table 3
The cooling performance comparisons in Cases 1, 5 and 6.

Parameter Case 1 Case 5 Case 6 Case 7 Case 8

Air mass flow rate (kg/s) 108.69 109.63 93.98 96.55 118.07
Decreased temperature of the cooling water (°C) 12.60 12.66 11.24 11.47 13.47
Air face velocity at the heat exchanger (m/s) 0.84 0.85 0.73 0.75 0.92
Air temperature at the heat exchanger exit (°C) 42.45 42.44 42.68 42.64 42.33
Overall UA based on the face area of the heat exchanger 758.13 766.16 671.82 687.24 823.19
Total air jet mass flow rate (kg/s) 16.5 16.5 16.5 12.4 24.2
Percentage increased in air mass flow, based on Case 0 52.27% 53.59% 31.66% 35.26% 65.40%
Percentage increased in heat transfer rate, based on Case 0 38.73% 39.19% 23.70% 26.29% 48.31%
Leverage factor of vortex 2.26 2.32 1.37 2.03 1.93

504
Y. Lu et al. Applied Energy 217 (2018) 496–508

1
not “sucked in” and “injected out” by any nozzle, although the line thus can be quantitatively expressed in the form of 2 ερ2 vθ2 . Here, ε is a
should be spiral instead of straight in its upper part as roughly illu- factor relating to the characteristic of the turbulent vortex including vθ
strated in Fig. 9. This difference suggests that the overall effect of the and the swirling ratio [44], which is out of the scope of this paper.
introduced jet is to add an extra momentum to the original tower air- Now combine Eqs. (15) and (19)–(22). Also, note that the terms
ΔM 1
flow at the nozzle plane. Suppose each jet nozzle injects an airstream − A θ and − 2 ρ2 vθ2 on each side of Eq. (21) can cancel out as the dynamic
inward at a speed vJ, and AJn is the sum of all nozzle areas—the area pressure due to the tangential velocity is only caused by the tangential
perpendicular to jet velocity. The total air jet mass flow rate from all momentum flux of the jet the plume obtains, but this does not mean the
eight nozzles, maJ , is plume’s angular (tangential) momentum disappears. We can now drive
the modified Draft Equation below:
maJ = ρI vJ AJn (18)
ρI vJ AJn ε K + 1−Kto 2
The initial momentum of the jet is then maJ vJ upon leaving the (ρO −ρI ) gH + ηz vzJ + ρ2 vθ2 ≅ ρI vz
A 2 2 (23)
nozzles with three components in the cylindrical coordinate system
written as: The right hand side of the equation remains similar to the one of Eq.
(15). The additional terms on the left hand side suggests that, when the
⎧ MrJ = vrJ maJ = maJ cosacosb vJ swirling plume is generated by the air jet in a natural draft cooling
MθJ = vθJ maJ = maJ sinacosb vJ tower, the increase in the “draft” (i.e. the left hand side of the Draft
⎨
⎩ MzJ = vzJ maJ = maJ sinb vJ (19) Equation) is attributed to two effects: the vertical momentum flux of the
ρ v A
jet the plume flow absorbs, namely the terms ηz vzJ I JA Jn , and the extra
where vrJ , vθJ and vzJ are jet speeds in radial, tangential and vertical
pressure change above the cooling tower outlet caused by plume vortex
(axial) directions, as indicated in Fig. 2. ε
ρ v 2 . The former directly influences the airflow’s dynamic pressure,
2 2 θ
The jet merges with the original airflow in the way which is es-
while the latter augments the static pressure though it is expressed in a
sentially a mixing of two streams with different speeds. The mixing
dynamic-pressure form. However, this change in static pressure is es-
occurs over a length, away from the nozzles, in which momentum is
sentially related to the tangential momentum flux of the jet, namely
exchanged. During the transient process, a part of or all initial jet ΔM
− Aθ .
momentum is converted into internal energy, or in other words, simply
It is the increase of the left hand side that causes the overall upward
lost. The remaining jet momentum is then added to the main airflow of
air flow rate vz to increase on the other side of Eq. (23). The above
the tower, and we can use ΔM to denote this remaining momentum.
comparisons in the CFD results for Cases 1–4 suggest that the tangential
Each component of the initial jet momentum in Eq. (19) has a different
momentum of the air jet is a dominant factor in improving the cooling
contribution to ΔM. The radial momentum MrJ causes a tendency of air
tower performance, as the pure vertical air jet increase the tower flow
moving towards the centre, which, obviously, will be quickly lost. The
rate only a little. This can be explained as that between the two extra
tangential momentum MθJ causes the air to rotate about the cooling ε
terms in Eq. (23), 2 ρ2 vθ2 is probably much larger.
tower axis—a key factor in generating the vortex. The vertical com-
The angular momentum of the swirling plume upon leaving the
ponent MzJ , on the other hand, increases the total upward flow mo-
tower outlet plane has a tendency to conserve. However, because of the
mentum. Then, it can be proposed that:
viscous resistance of the surrounding air, the momentum gradually
dissipates at some distance above the cooling tower. This leaves a
⎧ ΔMr = 0
ΔMθ = ηθ MθJ = ηθ vθJ maJ physical meaning for the aforementioned equivalent extended cooling
⎨ tower draft height ΔH . Mathematically, Eq. (23) can be simplified and
ΔMz = ηz MzJ = ηz vzJ maJ (20)
⎩
rearranged to the following form
where ηθ and ηz are the proportion factors ranging from 0 to 1 in tan- K + 1−Kto 2
gential and vertical directions, respectively. They measure the extent of (ρO −ρI ) g (H + ΔH ) ≅ ρI vz
2 (24)
“non-conservation of momentum” when the jet airstream merges with
the cooling tower airflow. They are essentially determined by vθj and vzj , where the invisible extended tower height ΔH is derived from the two
respectively, together with Ajn. additional terms of Eq. (23). For convenience, we may simply write ΔH
The airflow in the cooling tower is no longer purely caused by as a function of the air jet characteristic parameters, i.e.
natural draft, but a combination of natural draft and forced convection. ΔH = f (ηz ,vzJ ,AJn ,vJ ,ε ) or even ΔH = f (a,b,vJ ,AJn ) , by assuming the factor
Consequently, the original conservation of mechanical energy Eq. (12) ε is also a function of the air jet characteristic parameters.
is not valid, instead, it should read: Therefore, the improvement in the draft of the NDDCT is de-
termined by the jet speed, direction, and the jet nozzle size. This ar-
1 1 ΔMθ + ΔMz gument is supported by the CFD result comparison of Case 1 with Cases
P1−⎛P2 + ρ2 vz2 + ρ2 vθ2 + ρ2 gz 2⎞ = − + ΣHL
⎝ 2 2 ⎠ A (21) 5–8. For instances, increasing angle a from 60° to 70° while keep the
Note that all the quantities in the equation such as vz and vθ are angle b fixed at 30° has a better outcome, because the tangential mo-
representing their mean values over the tower cross-section. The term mentum of the jet MθJ = maJ sinacosbvJ is increased. However, in-
ΔM + ΔM
− θ A z on the right hand side is the momentum flux the plume flow creasing angle b from 30° to 45° while keep the angle b adversely affects
obtains from the jet streams and it is an averaged quantity as well. ΔH , as it increases MzJ but decreases MθJ . This further supports the
According to the analysis on the static pressure of the cooling tower claim that the jet tangential momentum that the swirling plume flow
with the swirling plume in Figs. 7 and 8, P2 is no longer slightly smaller absorbs is a critical factor. Furthermore, the jet mass flow rate is po-
than P3. Instead, an additional term should be introduced in Eq. (14) to sitively related to ΔH .
reflect the extra decrease of the static pressure above the cooling tower
outlet resulted by the swirling plume—ΔP : 4.2. Energy consumption of the air jet-induced swirling plume in NDDCT
α
P2 = P3− Kto ρ2 vz2−ΔP In this section, the power consumed on the NDDCT with air jet-
2 (22)
induced swirling plume in configuration of Case 5, the best case, is
The reason is that, because of the rotational motion of the air about estimated and compared against that of a conventional fan-forced
a fixed axis, any element of the flow field is exposed to an outward cooling tower/cooler with a same cooling capacity. The calculation of
pressure gradient due to the radial acceleration. As ΔP is expected to be fan power is based on basic fan theories and some simplified assump-
primarily determined by the tangential velocity of the vortex, its value tions.

505
Y. Lu et al. Applied Energy 217 (2018) 496–508

Hot source at 700 °C

The mechanical cooling tower is assumed to use exactly the same
horizontally-arranged heat exchanger bundles and two fans with a ty- Heater 2 Heater 1
pical diameter of 3.4 m [45] above the bundles instead of the 15 m-tall
tower shell. The fans provide an identical air flow rate through the heat
Recuperator Turbine
exchangers under the same ambient condition as in Case 5.
By theory, the shaft power input WF to fans or blowers is calculated
through total efficiency ηFt , i.e. Compressor
Cooling
VF ΔPFt tower
WF =
ηFt (26) Pre-cooler

where VF and ΔPFt are the volumetric flow rate and total pressure of the
fan, respectively. According to the definition of fan total pressure, one
Hot fluid Supercritical CO2 Cooling water
has:
Fig. 10. A schematic drawing of the split recuperated sCO2 Brayton cycle indirectly
1 1
VF ΔPFt = VF ⎛ΔPFs + ρa vF2 ⎞ = ΔPFs VF + ma vF2 cooled by the NDDCT with swirling plume.
⎝ 2 ⎠ 2 (27)
where ΔPFs is the static pressure of the fan, and vF is the air speed at the 1 MWe) [34]. Therefore, the enhancement of the heat dumping capacity
fan exit. It is difficult to estimate the static pressure of a single fan in the cooling tower with the proposed air jet-induced swirling plume
without knowing specific fan characteristics. However, in a fan system, can be evaluated, coupled with a performance calculation of an entire
the pressure is always determined by the total pressure resistance power conversion cycle for the CST plant [46].
throughout the system. According to Table 2, the air flow rate and the In this study, a modelling methodology similar to the ones in-
pressure drop across the heat exchangers are maintained at 109.63 kg/s troduced in [35] was adopted to establish a split recuperated super-
and 9.0 Pa, respectively. If we ignore all the pressure resistances other critical CO2 (sCO2) Brayton cycle model, with all the components illu-
than the one due to the heat exchangers [34], the useful work done by strated in Fig. 10. Assumptions were made to some of the key
the fans consists of the pressure energy that is used to overcome the component parameters as constraints or initial conditions of the cycle
pressure resistance of the heat exchangers and the kinetic energy at the model, as shown in Table 4. By adjusting the flow rates of the sCO2 and
two fan exits. Therefore, Eq. (27) becomes: hot source fluid, the cycle output a net power of 548.9 kWe with the
ΔPr ma 109.63 109.63 cool end (cooling tower) performance matching the result in the design
2
VF ΔPFt = + maF vaF = 9.0 × + point, i.e. the decreased cooling water temperature of 12.7 °C at the
ρa 1.145 2
2
flow rate of 15.5 kg/s and the ambient temperature of 30 °C.
109.63/2 With the hot source flow rate fixed, the temperature of the returning
×⎛⎜
2
⎞ = 2.39 kW
⎟

⎝ 1.145 × π × (3.4/2) ⎠ (28) water from the cooling tower was varied as per the cooling perfor-
mances derived from the results of Cases 1–9. Meanwhile, the cooling
where maF and vaF are the air mass flow rate and the speed in a single
water mass flow rate was also adjusted to maintain a constant tem-
fan. Then, the power consumption of the fans is
perature of the water leaving the pre-cooler, which was consistent with
2.39 the boundary condition of the aforementioned CFD heat exchanger
WF = kW
ηFt (29) model. The temperature changes in the cooling water resulted in a
series of consequent temperature and pressure changes within the cycle
On the other hand, for the air jet nozzles, the fans or blowers used in
so that the net power output varied accordingly. Fig. 11 depicts the net
the nozzles are assumed to work without any obstructions in order to
power output of the cycle for the design point and in Cases 0, 1, and 5.
keep the comparison fair. The pressure loss due to the fans themselves
The figure shows that the increase of the ambient temperature from
and ducts as well as nozzles are usually quite small and thus are also
30 °C (Design point) to 35 °C (Case 0) causes the net power output to
ignored here. Therefore, the air kinetic energy is the only useful work.
drop by 27.3 kWe, or 4.98%. In Cases 1 and 5, the cooling performance
The power consumption is then
of the NDDCT is enhanced by the jet-induced swirling plume at the
1 1
m v2
2 a J 2
× 16.5 × 202 3.30 higher ambient temperature. As a result, their cycle power outputs re-
WJ = = = kW cover to the level nearly same as that of the design point. The energy
ηFt ηFt ηFt (30)
consumption of the jet for Cases 1 and 5, on the other hand, is much
Obviously, the two power consumptions in Eqs. (29) and (30) rely smaller than the regained power output—around 1/7, according to the
on the total fan efficiencies. Now, if we assume these efficiencies are estimation in the previous section.
same, for example 80%, the total powers consumed in the vortex-en- The overall energy performance of the swirling plume cooling tower
hanced NDDCT and the fan-induced cooling tower are 4.1 kW and on the power cycle can be embodied in the consideration for long-term
3.0 kW, respectively. operation, e.g. one year period. Suppose the time when ambient
Although the power consumption in jet-induced swirling plume
NDCCT is higher, the two are still comparable. However, the huge
advantage of the former over the latter lies in life-long services. This is Table 4
Key component parameters in the sCO2 power cycle model.
because the vortex enhancement is in operation for short-term only.
The nozzle jet is only turned on in a small part of time over a year when Component parameter Value
the ambient temperature is too high for the NDDCT to perform as de-
signed; whereas the fan-based cooling tower relies on fans running all Heat source fluid temperature 700 °C
Turbine pressure ratio 3.1
the time.
Compressor inlet/outlet pressure @ design point 6.43 MPa/20 MPa
Turbine efficiency 85%
4.3. Effect of the swirling plume cooling tower on energy performance of Compressor efficiency 80%
power cycle Pressure loss at each heat exchanger 3%
Pinch point temperature difference at the recuperator 5 °C
Cooling water outlet temperature leaving the pre-cooler 51.5 °C
The prototype of the 20 m-tall NDDCT model was designed for an Heat exchange area of the pre-cooler 40 m2
expected supercritical CO2 cycle-based small CST power plant (up to

506
Y. Lu et al. Applied Energy 217 (2018) 496–508

560 The findings of the study point out that the jet-induced swirling
plume concept is potentially feasible in real applications on natural
Net power output of the cycle

550 draft cooling towers, either wet- or dry-type, to recover their cooling
capacity at high ambient temperatures. Although the power drawn by
540 the system is still comparable to that of a fan-forced cooler working
under the exactly same condition, the system would run only during
(kWe)

530 temperature extremes. Overall, the additional energy usage is relatively

small, leading to a low average power consumption. Besides, the
520
method has no influence to the NDCTs in normal pure natural con-
vection operations.
510
By identifying the critical factor in the vortex effects and introdu-
cing the leverage factor, the work suggest that, for a given NDCT, there
500
Design point Case 0 Case 1 Case 5 exists optimal combination/s among the air jet speed, direction and the
nozzle size. Other influencing parameters may also exist, including the
Fig. 11. The net power output of the cycle for the design point and in Cases 0, 1, and 5.
location of the air jets.

temperate is above the designed 30 °C accounts for 5% of a year, and Acknowledgements

the average temperate for this period is 35 °C. For the above super-
critical CO2-based Brayton cycle, the swirling plume concept would This research was performed as part of the Australian Solar Thermal
help to recover a net total of 10,088 kWh output electricity, compared Research Initiative (ASTRI), a project supported by the Australian
to the cycle with a normal 20 m-tall NDDCT. If the aforementioned Government, through the Australian Renewable Energy Agency
mechanical cooling tower/cooler is taken into the comparison, this (ARENA).
concept would save roughly 24,500 kWh parasitic electricity, by as-
suming the fan power is fixed at 3.0 kW all the time. References

5. Conclusions [1] Boulay RB, Cerha MJ, Massoudi M. Dry and hybrid condenser cooling design to
maximize operating income. In: ASME 2005 Power Conference, ASME, Chicago,
The heat dump performance of natural draft cooling towers Illinois, USA; 2005. p. 167–75.
[2] Stabat P, Marchio D. Simplified model for indirect-contact evaporative cooling-
(NDCTs) is one of the crucial factors affecting the overall power con- tower behaviour. Appl Energy 2004;78:433–51.
version (heat-to-electricity) efficiency in concentrated solar or coal- [3] Fisenko SP, Brin AA, Petruchik AI. Evaporative cooling of water in a mechanical
fired thermal power plants. To boost the energy efficiency of these draft cooling tower. Int J Heat Mass Transf 2004;47:165–77.
[4] Fisenko SP, Petruchik AI, Solodukhin AD. Evaporative cooling of water in a natural
power plants, a new concept of cooling performance enhancement in draft cooling tower. Int J Heat Mass Transf 2002;45:4683–94.
NDCTs against high ambient temperatures has been proposed by the [5] He S, Guan Z, Gurgenci H, Hooman K, Lu Y, Alkhedhair AM. Experimental study of
authors. The basic idea of the concept is to create high-swirling ratio film media used for evaporative pre-cooling of air. Energy Convers Manage
2014;87:874–84.
plume vortex in and above the towers by introducing a number of [6] Sun Y, Guan Z, Gurgenci H, Hooman K, Li X, Xia L. Investigation on the influence of
blower-powered air jets at certain locations. injection direction on the spray cooling performance in natural draft dry cooling
In this paper, using a 20 m-tall natural draft dry cooling tower tower. Int J Heat Mass Transf 2017;110:113–31.
[7] Alkhedhair A, Jahn I, Gurgenci H, Guan Z, He S, Lu Y. Numerical simulation of
(NDDCT) model, the effectiveness of the conceptual method on the
water spray in natural draft dry cooling towers with a new nozzle representation
cooling enhancement has been verified numerically and its working approach. Appl Therm Eng 2016;98:924–35.
mechanism has been initiatorily analysed. The benefit of using such a [8] Eugene Grindle JC. Roger Lawson improving natural draft cooling tower perfor-
mance with heat injection. In: 2002 International Joint Power Generation
swirling plume NDDCT to power cycles as well as its power consump-
Conference, ASME, Scottsdale, Arizona, USA; 2002.
tion have been investigated through simplified analytical models. The [9] Zou Z, Guan Z, Gurgenci H, Lu Y. Solar enhanced natural draft dry cooling tower for
study comes to the following conclusions. geothermal power applications. Sol Energy 2012;86:2686–94.
[10] Smrekar J, Oman J, Širok B. Improving the efficiency of natural draft cooling
towers. Energy Convers Manage 2006;47:1086–100.
1. The jet-induced swirling plume is able to improve the cooling per- [11] Li X, Xia L, Gurgenci H, Guan Z. Performance enhancement for the natural draft dry
formance of NDCTs. Through numerical modelling, we found that cooling tower under crosswind condition by optimizing the water distribution. Int J
the proposed plume vortex enhances the total air flow rate and the Heat Mass Transf 2017;107:271–80.
[12] Zhao Y, Long G, Sun F, Li Y, Zhang C, Liu J. Effect mechanism of air deflectors on
decreased cooling water temperature of the 20 m-tall NDDCT by at the cooling performance of dry cooling tower with vertical delta radiators under
least 53.6% and 3.57 °C (39.3%) respectively, at 35 °C ambient crosswind. Energy Convers Manage 2015;93:321–31.
temperature. The enhancement enables the cooling tower recover its [13] Zhao Y, Sun F, Li Y, Long G, Yang Z. Numerical study on the cooling performance of
natural draft dry cooling tower with vertical delta radiators under constant heat
performance to the level nearly same to that at 30 °C ambient load. Appl Energy 2015;149:225–37.
temperature. [14] Zhao YB, Long G, Sun F, Li Y, Zhang C. Numerical study on the cooling performance
2. The swirling plume NDDCT increases the net power output of a half of dry cooling tower with vertical two-pass column radiators under crosswind. Appl
Therm Eng 2015;75:1106–17.
megawatt-scale sCO2-based CST power cycle by 4.98% at the at-
[15] Goodarzi M, Ramezanpour R. Alternative geometry for cylindrical natural draft
mospheric temperature of 35 °C. The energy consumption for cooling tower with higher cooling efficiency under crosswind condition. Energy
creating such a benefit only accounts for around 1/7 of the regained Convers Manage 2014;77:243–9.
[16] Wang W, Zhang H, Liu P, Li Z, Lv J, Ni W. The cooling performance of a natural
electricity.
draft dry cooling tower under crosswind and an enclosure approach to cooling ef-
3. The concept works for the reason that the swirling plume creates an ficiency enhancement. Appl Energy 2017;186:336–46.
equivalent extra draft height on top of the tower, ΔH , which is at- [17] Wang W, Lyu J, Zhang H, Liu Q, Yue G, Ni W. A performance enhancement of a
tributed to two vortical effects. ΔH is then proposed to be func- natural draft dry cooling tower in crosswind via inlet flow field reconstruction.
Energy Build 2018;164:121–30.
tionally related to the air jet speed, the direction (both the azimuthal [18] Ma H, Si F, Kong Y, Zhu K, Yan W. Wind-break walls with optimized setting angles
pitching angles), and the nozzle size. The study compared 9 com- for natural draft dry cooling tower with vertical radiators. Appl Therm Eng
bination cases of the above parameters, and found that the angular 2017;112:326–39.
[19] Wang Q, Zhu J, Lu X. Numerical simulation of heat transfer process in solar en-
momentum of the swirling plume is the critical factor in determining hanced natural draft dry cooling tower with radiation model. Appl Therm Eng
the effect of the swirling plume. 2017;114:977–83.
[20] Kong Y, Wang W, Yang L, Du X, Yang Y. A novel natural draft dry cooling system

507
Y. Lu et al. Applied Energy 217 (2018) 496–508

with bilaterally arranged air-cooled heat exchanger. Int J Therm Sci cooling tower. Appl Therm Eng 2016;105:1013–20.
2017;112:318–34. [35] Li X, Duniam S, Gurgenci H, Guan Z, Veeraragavan A. Full scale experimental study
[21] Chen L, Yang L, Du X, Yang Y. Performance improvement of natural draft dry of a small natural draft dry cooling tower for concentrating solar thermal power
cooling system by interior and exterior windbreaker configurations. Int J Heat Mass plant. Appl Energy 2017;193:15–27.
Transf 2016;96:42–63. [36] Lu Y, Gurgenci H, Guan Z, He S. The influence of windbreak wall orientation on the
[22] Alavi SR, Rahmati M. Experimental investigation on thermal performance of nat- cooling performance of small natural draft dry cooling towers. Int J Heat Mass
ural draft wet cooling towers employing an innovative wind-creator setup. Energy Transf 2014;79:1059–69.
Convers Manage 2016;122:504–14. [37] Lu Y, Guan Z, Gurgenci H, Alkhedhair A, He S. Experimental investigation into the
[23] Liao HT, Yang LJ, Wu XP, Du XZ, Yang YP. Impacts of tower spacing on thermo-flow positive effects of a tri-blade-like windbreak wall on small size natural draft dry
characteristics of natural draft dry cooling system. Int J Therm Sci cooling towers. Appl Therm Eng 2016;105:1000–12.
2016;102:168–84. [38] Lu Y, Guan Z, Gurgenci H, Zou Z. Windbreak walls reverse the negative effect of
[24] Moore FK. On the minimum size of large dry cooling towers with combined me- crosswind in short natural draft dry cooling towers into a performance enhance-
chanical and natural draft. J Heat Transf 1973;95:383–9. ment. Int J Heat Mass Transf 2013;63:162–70.
[25] Kroger DG. Air-cooled heat exchangers and cooling towers. Tulsa, Okl, USA: [39] Lu Y, Guan Z, Gurgenci H, Hooman K, He S, Bharathan D. Experimental study of
Pennwell Corp; 2004. crosswind effects on the performance of small cylindrical natural draft dry cooling
[26] Gasparini G, Barbieri G. Performance augmentation of natural draft cooling towers. towers. Energy Convers Manage 2015;91:238–48.
In: Google Patents; 2006. [40] Li X, Gurgenci H, Guan Z, Wang X, Duniam S. Measurements of crosswind influence
[27] Bosman P. Fan-assisted wet coolong tower and method of reducing liquid loss. In: on a natural draft dry cooling tower for a solar thermal power plant. Appl Energy
Google Patents; 2005. 2017;206:1169–83.
[28] Schreiber H. Fan cooling tower design and method. In: Google Patents; 2014. [41] Meyyappan M, Schwarz M, Perry J. Modelling of swirl jet flows. In: 1st
[29] Gerald RAW, Stillman I. Confined Vortex Cooling Tower, in USA; 1983. International Conference on CFD in the Mineral & Metal Processing and Power
[30] McAllister Jr JE, Aiken SC. Vortex-augmented cooling tower-windmill combination, Generation Industries, Melbourne, Australia; 1997.
in, the United States Department of Energy, Washington, D.C., USA; 1985. [42] A. Inc., FLUENT User's Guide, vol. 2011, ANSYS Inc., New Hampshire, USA, 2011.
[31] Kashania MMH, Dobregob KV. Heat and mass transfer in the over-shower zone of a [43] Henderson-Sellers B. The zone of flow establishment for plumes with significant
cooling tower with flow rotation. J Eng Phys Thermophys 2013;86:1490–9. buoyancy. Appl Math Model 1983;7:395–8.
[32] Kashani MMH, Dobrego KV. Influence of flow rotation within a cooling tower on [44] Klimenko AY. Strong swirl approximation and intensive vortices in the atmosphere.
the aerodynamic interaction with crosswind flow. J Eng Phys Thermophys J Fluid Mech 2014;738:268–98.
2014;87:385–93. [45] Lu Y, Guan Z, Hooman K, Parulekar PS. An investigation on cooling performance of
[33] Dobrego KV, Davydenko VF, Koznacheev IA. Use of oriented spray nozzles to set the air-cooled heat exchangers used in coal seam gas production. Heat Transf. Eng
vapor-air flow in rotary motion in the superspray space of the evaporative chimney- 2016.
type tower. J Eng Phys Thermophys 2016;89:157–66. [46] Deng H, Boehm RF. An estimation of the performance limits and improvement of
[34] Li X, Guan Z, Gurgenci H, Lu Y, He S. Simulation of the UQ Gatton natural draft dry dry cooling on trough solar thermal plants. Appl Energy 2011;88:216–23.

508